The aim of this report is to assist Donkey Games in deciding what their next game project should be. We will determine what types of games typically perform well in terms of sales, and ultimately advise Donkey Games on potential next steps. Many different factors will be taken into consideration, including platform as well as game genre.
The data used has been sourced from VGChartz, a video game industry news website with a focus on console and game sales. Four different CSV files will be used to complete the report.
The data contains records of game sales between the years 1970 and 2019, across many different platforms and game genres. There is a high number of missing values which will need to be addressed, and in terms of bias certain gaming platforms for example are more represented than others. Overall however, there is a lot of data available from which to extract potentially valuable insights.
The data does not contain anything that could be considered PII (potentially identifiable information). However, there are links to box art within the datasets, so copyright laws might be a valid concern.
After an initial data cleaning phase, the resulting joined dataset was used to create an explanatory multiple linear regression model. The aim was to attempt to explain which factors might impact a game's number of global sales. Multiple models were created iteratively, leading to different insights. The models created are available for review in individual notebooks within the notebooks folder, and the insights extracted are summarised below.
The main insights gained through modelling the data available to us are as follow:
Although the process led to some valuable insights, there is room to build more robust models once more varied data is available.
First, let us take a look at sales across all gaming platforms represented in our data.
Note: All visualisations included in this report were created with Plotly Express and are interactive. You will be able to hover over columns to view individual figures, zoom in and out if needed, and restrict the visualisations to certain parameters as needed by clicking on the legend to the right of each plot.
This plot allows us to view the overall cumulative sales numbers across all game genres in our data.
In order to gain a better understanding, we can also look at the mean number of sales by game for each genre.
This plot confirms one of our modelling findings, games that perform well in terms of sales tend to be well reviewed. However, good reviews do not guarantee high sales numbers.
Although some aspects of this bar plot are hard to interpret, it provides a visualisation of the scale of the video game industry. We also have confirmation of some of our modelling findings; shooters and action games appear to be quite popular in recent years.
Here we have plotted the global sales for the 10 best performing games in our data. As we can see, there is a wide spread of different genres, and 8 of the games were released on Nintendo systems. The issue for us is that Nintendo produce most of their own games in-house, so we will mostly be focusing on other platforms.
For the following sections of the report, we will focus on PC games, as well as the two major competing console platforms, Sony's PlayStation 4 and Microsoft's Xbox One.
The following dashboard was designed to be used in conjunction with this report and will allow you to further explore individual game genres as needed:
Here we can see which game genres have the highest number of cumulative sales on PC.
This mean global sales plot, along with the previous plot, can help us narrow down the game genres to focus on. For the following plot we will take a look at five popular genres:
Here we can see an indication of trends across five of the most popular PC game genres. This will allow us to make an informed decision as to which genre or genres to consider.
Note: Click on individual genres within the legend to the right of the plot to remove or add genres as needed.
We can take a closer look at simulation games for example, and gain an understanding of which publishers it might be beneficial to work with based on their prior track record.
All these visualisations combined should allow us to decide which direction to follow if we consider developing a PC title.
Following the same methodology as above, we can now take a look at PS4 titles.
Based on the first two plots, we can restrict our next visualisation to the following game genres:
Looking at the above trends, it seems that Action-Adventure titles in particular are experiencing an upswing. We can take a closer look at publisher recommendations in the following visualisation.
Note: Click on individual genres within the legend to the right of the plot to remove or add genres as needed.
Here we can see that Rockstar Games typically sell a high number of copies within the Action-Adventure genre. However, as they tend to develop their titles in-house they may not be an option for us. It may be worth exploring other options in this case, such as 2K Games for example.
Finally, following the same methodology once again, we can explore Xbox One titles.
Based on our first two visualisations, we can restrict our next step to the following game genres:
Based on this visualisation we might decide to look into developing a Racing title. Although they might not have the potential to sell as many copies as games within other genres, their consistency could indicate a lesser risk for us.
Note: Click on individual genres within the legend to the right of the plot to remove or add genres as needed.
It seems that Microsoft Studios would be a good choice for a potential partner should we decide to develop a Racing title on Xbox One.
Based upon this report the following recommendations can be made:
Additionally there are some limitations we need to be aware of. Our data does not contain any revenue figures, quarterly sales numbers, or game budgets.
Without any such figures it is hard to determine what type of game might ultimately result in the desired return on investment.
Thank you for reading!